Final Report on GR/M34638/01

"A Magnetic Force Microscope for the Study of Magnetic Nanostructures and Advanced Magnetic Materials"
1. Background

Magnetic force microscopy (MFM) has become, with the increased availability of commercial instruments, one of the most commonly used magnetic imaging techniques over the last 10 years. Whilst the technique is used routinely to observe many different samples, a quantitative understanding of contrast obtained in the images has developed rather slowly. In this project funding was provided to obtain a state-of-the-art magnetic force microscope (Digital Instruments Dimension 3100) to complement the existing world class transmission electron microscopy (TEM) imaging currently being implemented at the University of Glasgow. The extensive experience of quantitative magnetic imaging at Glasgow using electron microscopy (Lorentz microscopy) allowed well characterised systems to be studied using the MFM to help develop quantitative image interpretation in this mode. Furthermore MFM is well suited to study systems which are not easily imaged by TEM namely bulk samples or thin film samples on bulk substrates. We have used the MFM to look at a number of systems of this type e.g. recording heads and magnetic elements integrated in electrical devices. The instrument may also be used as an atomic force microscope (AFM) to investigate the detailed surface structure of materials. A number of projects are briefly described using this mode and include some with cross-disciplinary aspects. Details of the various projects are given in the next section and the main achievements of the project are summarised below:
· quantitative comparison of Lorentz microscopy and MFM images of model systems with those predicted by simulation software
· imaging of magnetic behaviour of as deposited and thermally annealed arrays of magnetic elements with variable spacing and size using TEM and MFM
· analysis of the form of the writing field distribution from recording heads and correlation with field reconstruction from Lorentz microscopy
· imaging of integrated magnetic/electrical devices and surface studies of radiation damage in particle physics detectors

2. Key Advances and Supporting Methodology

2.1. Magnetic Thin Film Element Arrays
Arrays of magnetic elements have been proposed for a number of applications, for example, high speed non-volatile magnetic memory, logic functions and computation. In all these applications, element interactions are very important, This is clearly undesirable for storage applications, but very necessary for logic or computation. The material used and the element shape affects the magnetisation state, switching field, etc. Various groups worldwide having studied shapes including rectangles, circles, triangles and pentagons. It is unclear exactly what range of behaviour might be expected as shape and coupling is varied. As part of a PhD project, work with the MFM has demonstrated that much richer behaviour may be encountered than might naively be expected. The magnetic elements were patterned by standard methods of high resolution electron beam lithography. The shapes were on the micron scale. Although somewhat larger than what would be expected for practical applications, the sizes used permitted ready imaging by MFM
The magnetised state of the as-deposited arrays was generally not the global ground state, but a higher energy (metastable) state. Characterisation of both states is important for storage systems. For example, hard disks are susceptible to thermal spikes and temperature transients may be sufficient to induce local modifications in the written information if the coupling to nearby 'bit' magnetisations is sufficiently strong. We investigated both the metastable and the global ground state. The ground state for the MFM samples was induced by annealing in a RTA (rapid thermal annealer) at >250ºC for a few minutes; the Curie temperature of the as-deposited permalloy is around 580ºC. Rectangular elements are the naive choice to exploit for storage applications. We used an aspect ratio of 4:1. Such elements support a number of possible flux closed states. Indeed, in the as-deposited array both flux closed and near-uniformly magnetised elements were seen. However, annealing induced uniform magnetisation through the whole array. The coupling between elements is due to dipole-dipole (magnetostatic) interactions and appears to be quite strong. Tapering the ends of a rectangle enhances shape anisotropy. Arrays of rhombic shapes were also patterned. These did not exhibit any tendency to flux close and the as-deposited state was spontaneously magnetised to its ground state without the need for annealing. Again strong magnetostatic coupling is observed.

 

Fig 1. MFM images of triangular elements showing (a) strong dipole coupling and (b) appearance of "superdomains"

 For triangular elements placed on a triangular lattice, a number of intricate interaction phenomena were observed. Combined MFM and TEM imaging showed that the triangles may be magnetised in vortex and non-vortex states. This was confirmed by micromagnetic simulations using the package LLG™. Arrays were imaged by MFM as a function of spacing before and after annealing. For large spacings the elements were all in vortex states. As the spacing was reduced, the effect observed depended on film thickness. For thick (>50 nm) and thin (<35 nm) films the elements remained in the vortex state. For intermediate film thicknesses, the vortices were expelled from the elements below a critical spacing. As the spacing was reduced further, superdomains of vortex states were nucleated, the circulations being random. For very close spacing, the vortex superdomains filled the entire array, and their circulations became aligned.
2.2. MFM tip characterisation and MFM and magnetic TEM image calculation
During the course of this project we were able to bring together imaging of simple structures using both Lorentz microscopy (Differential Phase Contrast - DPC) and MFM together with MFM tip field reconstructions made also using Lorentz microscopy. By applying these to a simple magnetic system (a thin film element) whose magnetic state can be simulated using advanced software packages we have been able to make considerable progress in quantification of the contrast observed in the MFM images. The software packages used are LLG™ and OOMMF (Object Oriented Micromagnetic Framework) [1, 2].
The imaging experiments and simulations were performed on a 20nm thick permalloy element with in-plane dimensions 2.0mmx1.0mm which is one of the standard problems in the field of micromagnetic calculations [3]. The magnetic state of the element we considered comprised a simple flux closure vortex structure with four domains, which was also observed in the experimental case. For a given magnetisation state we have developed software routines which can calculate the expected DPC and MFM images using a method based on Fourier transforms. These can be implemented in a commercial software package.
DPC images are maps of components of the induction integrated along the path of the electron beam in the electron microscope. The DPC images from this element resemble the magnetisation images as it is really the domain walls that are the main sources of stray field and the overall structure of the element can be considered as of the flux closure type. The experimental and simulated images are very closely matched as can be seen in Fig 2. The high spatial frequency signal variation is due to the differential scattering from the crystallites (5-10nm in size) which make up the film.
Fig.2 (a) Simulated DPC image of 20nm thick 2.0µm x 1.0µm element of permalloy. (b) Experimental DPC image of central portion of element. (c) Linetraces taken between points A and A' for images (a) and (b) as indicated from position in (a).
The above results confirm the validity of the simulation package in this case and the quantitative nature of DPC contrast for imaging the magnetic structure on the scale of <10nm.Image interpretation in MFM can be considered from different viewpoints. Perhaps the simplest is that it represents a form of charge microscopy [4]. In this interpretation the image represents a magnetic charge distribution (tip/sample) convoluted with a stray field or field gradient (sample/tip) dependent on whether the force or force gradient is being measured. In calculating or simulating MFM images it then becomes necessary to give a realistic representation for the MFM probe. Whilst point charges may be used for image calculations, these generally give images with superior resolution to those achieved in experiments. At Glasgow we are able to reconstruct the MFM tip field from DPC linetraces and tomographic reconstruction techniques [5]. From these measurements it is possible to assign an equivalent charge distribution associated with the tip which gives rise to the stray field for calculation purposes. This assumes the charge is associated with a plane which is perpendicular to the axis of the tip and contains the apex of the tip at the lift height. The extended charge distribution is determined from comparison with the field distribution of the tip reconstructed from DPC images of an axially magnetised standard Digital Instrument MESP tip. In this case the perturbative effect of the tip field on the sample is apparent as the Bloch line where the polarity of the Néel wall changes is very mobile and can be seen to have shifted to the junction of the 90o walls in Fig. 3(b). It should be noted that debris on he surface of the elements were a problem for this type of t imaging and in this case an element of slightly larger dimensions has been studied. However, comparison of the linetraces across the wall from the simulation and experiment are in excellent agreement as can be seen in Fig. 3(c)
Fig. 3 (a) Simulated MFM image of permalloy element of fig.2 taken at a lift height of 50nm for standard DI MESP probe. (b) Experimental MFM image of central portion of a similar sized element. (c) Linetraces taken across domain wall (from B to B') for images (a) and (b).
2.3. Recording heads
Our work on magnetic recording using the MFM has mainly been in association with Onstream, a company established to produce and develop multi-track recording on 8mm tape. Our studies involved both the Emboss heads, which are used to write servo information in the tape during manufacture and the data read/write heads, which operate on 8 parallel tracks. The initial study of the fields from these transducers was carried out by STEM imaging using the DPC method. Both types of head presented considerable problems due to their large physical size and method of construction (not however a problem with MFM). For the Emboss head, DPC imaging results, with only one of the two rows of 96 heads driven, indicated that there could be substantial stray field escaping from the heads of the undriven row; this would imply a very inefficient head design. However modelling suggested that the observed contrast could arise from side fields from the driven poles, but this was not confirmed until we were able to conduct MFM imaging studies; these also showed that the writing field variation from head to head in the array was very small. Although STEM DPC imaging can permit quantitative head field reconstruction by electron beam tomography [6], it is difficult to do this for distances <0.25mm from the pole surfaces and flying heights in state-of-art disk files are now substantially less than this. Hence we are investigating if it is possible to get suitable field data, eitherby MFM studies alone, or more probably in combination with STEM tomography on the same recording head. A precursor to this is an understanding of the MFM tip/head interaction and we have studied several tips ranging from magnetically soft to hard. The conclusion we can drawis that ideally the tips should be either
very soft or very hard, but neither state is simple to achieve in practice [7]. In our studies, image interpretation was complicated by the influence on the contrast of the permeability of the polepiece material. To overcome this we have developed an image subtraction method which uses either a combination of driven and undriven head images or of images obtained with drive currents of equal and opposite magnitude; an example using the former method is illustrated Fig. 4.
Fig. 4a (a) Original MFM image of Emboss head. (b) Subtracted MFM image (c) Linetraces from images (a) and (b) taken between points C and C'
2.4. Miscellaneous Magnetic Samples

2.4.1 Miniature electromagnet for localised field application
A micron-scale electromagnet is being developed within a PhD project with a nanometre scale pole piece. Characterisation of the electromagnet operation is done using the MFM. Evidence of stray fields from the pole piece influencing nanoelements adjacent to it has been found. The aim is to integrate the electromagnet with TEM membranes so that local fields may be applied in magnetisation studies of small elements. Typically in TEM studies a global field is applied across the entire specimen. In work with Seagate, a micro-electromagnet is under development so that, for example, individual elements may be pulsed with a field without that field impinging significantly on elements more distant from the pole piece. There is also interest in having small electromagnets that can apply the rather large fields required to write to high coercivity media.
2.4.3 Surface profiles of magnetic elements
For nanometre scale magnetic elements, edge definition is important to control switching uniformity. A number of lithographic routes are possible, but two have been exploited actively for sample preparation in Glasgow. The first is by electron-beam lithography and lift off; the second is film deposition followed by dry etching through a mask patterned by electron-beam lithography. The former is a well-used technique but suffers when used on thin substrates as there are problems with contact between the deposited metal and resist at the edges of elements. This results in unwanted metal flags at the edges, which give rise to large contrast in Lorentz microscopy which is non-magnetic in origin.
Fig. 5 Optical image of miniature electromagnet
The latter process is harder to implement due to problems of corrosion but results in a better edge profile. The AFM/MFM has been very useful for characterisation of the element edge profile to determine problems associated with each process and therefore to optimise production of the elements for TEM imaging.

2.4.4 Other Magnetic Activities
In addition to the work described above the MFM has also been used in other projects which are still in progress; (i) In a collaboration with the AFM group in the Department of Electronics & Electrical Engineering and the Magnetics group at Leeds University, some preliminary work has been carried out to study the stray field from perpendicular media. The ongoing plan is to write to these films using single turn nanometre scale loops fabricated on micromachined AFM cantilevers and to characterise the written bits by MFM or TEM. Some of the thicker perpendicular media films showed characteristic stripe domain structures in MFM. (ii) As part of a final year undergraduate project and a recently started PhD project some preliminary studies have been performed on a hybrid semiconductor-magnetic system comprising a nanoelement coupled to a Hall geometry defined by Schottky gates on a high mobility GaAs/AlGaAs heterostructure. Some features in the magnetoresistance reminiscent of the Aharonov-Bohm effect have been observed at dilution fridge temperatures. (iii) Finally some exploratory work has been undertaken to use magnetic shapes for defining magnetic logic operations. Some interesting results were obtained as part of an undergraduate final year project but these have to be taken further to extract their significance.

2.5. AFM Applications

2.5.1 Surface lateral superlattices
The transport of electrons in artificial periodic potentials has been an active area of condensed matter research over a number of years. Typically these potentials are defined in the 2 dimensional electron gas (2DEG) of a GaAs/AlGaAs heterostructure. Recent developments have focussed on 2D periodic potentials where a number of fundamental unresolved issues have recently been addressed through work at Glasgow [7]. Crucial to this has been the fabrication of sub-100 nm period 2D periodic structures with high fidelity. AFM imaging has been central to achieving the desired level of reproducibility in fabrication, particularly with regard to the etch depth during sample preparation. This depth is only a few nm and must be controlled with a few percent error.
Fig.6 Etched lattice structures for potential modulation of 2DEG. (a) Scanning electron microscope image. (b) AFM image.
2.5.2 Radiation damaged surfaces
In development of detector systems for future high energy physics experiments under the PPARC remit, there is interest in searching for semiconductor materials that are potentially more radiation resistant than silicon or GaAs. A study has been initiated on the suitability of SiC and GaN, focussing latterly on just SiC, as particle detection media. The AFM has been used to probe the surface of materials bombarded at CERN (24 GeV protons), to correlate surface structure with electrical performance. An increase in surface roughness was found, but electrical behaviour was not degraded significantly in pre-patterned diodes suggesting some degree of radiation resistance. For free surfaces, the effects measured were similar to those seen when semiconductors are exposed to a low energy plasma (<1 keV ions), indicative of some lattice damage.

2.5.3 Retinal electrode profiles
At the interface between the physical and life sciences there is interest in developing systems that couple to neural tissues and are able to measure large numbers of parallel neural spike trains. The interface is usually achieved by microfabricated electrode arrays, where the electrodes are platinised to reduce the contact impedance. A protective dielectric coating is used on the arrays and platinisation is carried out through vias defined by etching through to electrical tracks beneath. For uniformity in platinisation and contact impedance the vias should be identical. AFM measurements have shown that this is not always the case and that there may be some degree of variability in the profile of the vias.

3. Project Plan Review

The work progressed as described by the project plan review without any major changes in emphasis or impediments.

4. Research Impact and Benefits to Society

Through the course of the work described in this project emphasis has been placed on quantitative interpretation of MFM images and on the utility of comparison with results from simulation software. Although we have restricted ourselves at the moment to a non-perturbative approach to imaging we have been able to show that quantitative analysis is possible provided the MFM probe has been characterised suitably. The magnetic TEM imaging has been invaluable in supporting the results from the MFM. The global community of MFM users will benefit from the issues addressed in this study of quantitative MFM imaging. The research has also highlighted the complementary character of magnetic TEM and MFM imaging which are extremely powerful when used together.

A large and successful part of the project involved training of researchers (staff, postdoctorates, graduate and undergraduate students) to gain skills in acquiring and analysing images from the MFM. In some cases MFM imaging has become a significant part of their projects. The personnel involved are listed in the grant review form.

5. Explanation of Expenditure
 The project was funded under the JREI scheme with EPSRC contributing £52,997 and a further £46,000 from the University of Glasgow and industrial collaborators (Seagate Technologies, Ford Motor Company Ltd., Philips Research Laboratories). As this was a JREI award the full amount was spent only on the purchase of equipment. Drs McVitie and Rahman performed tests on the main instruments provided by the leading companies at the time i.e. Digital Instruments (DI) and Park Scientific. The Digital Instruments machines were found to be superior in performance and ease of use. Initially it had been intended to purchase the MultiMode instrument which accommodates smaller samples, however after negotiations with DI representatives we were able to secure the larger stage Dimension 3100 for £99k reduced from an original quoted price of £107k.

6. Further Research and Dissemination Activities

Although this project only ran for 2 years a number of significant papers have been published, accepted for publication or been submitted to major journals. Furthermore the results described in this report have been presented at major conferences and workshops. We have a strong collaboration with other UK groups and have a key role in the EPSRC Advanced Magnetics Program Network on Structural Characterisation and Imaging. Our experiences reported in this project have been well received in the latter. A proposal which utilises the MFM in the study of magnetic multilayers has already been submitted to EPSRC in June 2001. In addition it is expected that a further application will be made to EPSRC in late 2001 to investigate MFM imaging with in-situ magnetising capability and magnetisation reconstruction using improved magnetic TEM imaging and MFM techniques.

References

[1] MR Scheinfein, details at http://www.dancris.com/~llg/.
[2] MJ Donahue and DG Porter, details at http://math.nist.gov/oommf/.
[3] W Rave and A Hubert, IEEE Trans. Magn., 36, 3886-3899, (2001).
[4] A Hubert, W Rave and SL Tomlinson, Phys. Status Solidi B, 204, 817, (1997).
[5] R P Ferrier, S McVitie, A Gallagher and WAP Nicholson, IEEE Trans. Magn., 33, 4062, (1997).
[6] RP Ferrier and Y Liu, J. Magn. Magn. Mater., 175, 111, (1997).
[7] T Munro, Ph.D Thesis, University of Glasgow (2001).
[8] DE Grant, AR Long and JH Davies, Phys. Rev. B, 61, 13127 (2000).